X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/2a784fa3d4533d11e3227f868085dcee5a9870c6..1424e8e29bd54d3f713a38a834ff0c8ff5f463fb:/utils/setup.php diff --git a/utils/setup.php b/utils/setup.php index fc5b8e74..89177914 100755 --- a/utils/setup.php +++ b/utils/setup.php @@ -38,6 +38,7 @@ $aCMDOptions array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'), array('index-noanalyse', '', 0, 1, 0, 0, 'bool', 'Do not perform analyse operations during index (EXPERT)'), array('create-search-indices', '', 0, 1, 0, 0, 'bool', 'Create additional indices required for search and update'), + array('create-country-names', '', 0, 1, 0, 0, 'bool', 'Create default list of searchable country names'), array('drop', '', 0, 1, 0, 0, 'bool', 'Drop tables needed for updates, making the database readonly (EXPERIMENTAL)'), ); getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true); @@ -64,11 +65,11 @@ if ($aCMDResult['import-data'] || $aCMDResult['all']) { $iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1); if ($iInstances < 1) { $iInstances = 1; - echo "WARNING: resetting threads to $iInstances\n"; + warn("resetting threads to $iInstances"); } if ($iInstances > getProcessorCount()) { $iInstances = getProcessorCount(); - echo "WARNING: resetting threads to $iInstances\n"; + warn("resetting threads to $iInstances"); } // Assume we can steal all the cache memory in the box (unless told otherwise) @@ -82,7 +83,7 @@ $aDSNInfo = DB::parseDSN(CONST_Database_DSN); if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; if ($aCMDResult['create-db'] || $aCMDResult['all']) { - echo "Create DB\n"; + info("Create DB"); $bDidSomething = true; $oDB = DB::connect(CONST_Database_DSN, false); if (!PEAR::isError($oDB)) { @@ -92,9 +93,8 @@ if ($aCMDResult['create-db'] || $aCMDResult['all']) { } if ($aCMDResult['setup-db'] || $aCMDResult['all']) { - echo "Setup DB\n"; + info("Setup DB"); $bDidSomething = true; - // TODO: path detection, detection memory, etc. $oDB =& getDB(); @@ -115,25 +115,55 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) { if ($iNumFunc == 0) { pgsqlRunScript("create function hstore_to_json(dummy hstore) returns text AS 'select null::text' language sql immutable"); - echo "WARNING: Postgresql is too old. extratags and namedetails API not available."; + warn('Postgresql is too old. extratags and namedetails API not available.'); } $fPostgisVersion = getPostgisVersion($oDB); echo 'Postgis version found: '.$fPostgisVersion."\n"; if ($fPostgisVersion < 2.1) { - // Function was renamed in 2.1 and throws an annoying deprecation warning + // Functions were renamed in 2.1 and throw an annoying deprecation warning pgsqlRunScript('ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint'); + pgsqlRunScript('ALTER FUNCTION ST_Line_Locate_Point(geometry, geometry) RENAME TO ST_LineLocatePoint'); + } + if ($fPostgisVersion < 2.2) { + pgsqlRunScript('ALTER FUNCTION ST_Distance_Spheroid(geometry, geometry, spheroid) RENAME TO ST_DistanceSpheroid'); + } + + $i = chksql($oDB->getOne("select count(*) from pg_user where usename = '".CONST_Database_Web_User."'")); + if ($i == 0) { + echo "\nERROR: Web user '".CONST_Database_Web_User."' does not exist. Create it with:\n"; + echo "\n createuser ".CONST_Database_Web_User."\n\n"; + exit(1); + } + + // Try accessing the C module, so we know early if something is wrong + // and can simply error out. + $sSQL = "CREATE FUNCTION nominatim_test_import_func(text) RETURNS text AS '"; + $sSQL .= CONST_InstallPath."/module/nominatim.so', 'transliteration' LANGUAGE c IMMUTABLE STRICT"; + $sSQL .= ';DROP FUNCTION nominatim_test_import_func(text);'; + $oResult = $oDB->query($sSQL); + + if (PEAR::isError($oResult)) { + echo "\nERROR: Failed to load nominatim module. Reason:\n"; + echo $oResult->userinfo."\n\n"; + exit(1); + } + + if (!file_exists(CONST_ExtraDataPath.'/country_osm_grid.sql.gz')) { + echo "Error: you need to download the country_osm_grid first:"; + echo "\n wget -O ".CONST_ExtraDataPath."/country_osm_grid.sql.gz http://www.nominatim.org/data/country_grid.sql.gz\n"; + exit(1); } pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql'); pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql'); - pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql'); + pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql.gz'); pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql'); if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) { pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz'); } else { - echo "WARNING: external UK postcode table not found.\n"; + warn('external UK postcode table not found.'); } if (CONST_Use_Extra_US_Postcodes) { pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql'); @@ -152,12 +182,13 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) { } if ($aCMDResult['import-data'] || $aCMDResult['all']) { - echo "Import\n"; + info('Import data'); $bDidSomething = true; $osm2pgsql = CONST_Osm2pgsql_Binary; if (!file_exists($osm2pgsql)) { - echo "Please download and build osm2pgsql.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n"; + echo "Check CONST_Osm2pgsql_Binary in your local settings file.\n"; + echo "Normally you should not need to set this manually.\n"; fail("osm2pgsql not found in '$osm2pgsql'"); } @@ -179,22 +210,24 @@ if ($aCMDResult['import-data'] || $aCMDResult['all']) { passthruCheckReturn($osm2pgsql); $oDB =& getDB(); - if (!chksql($oDB->getRow('select * from place limit 1'))) { + if (!$aCMDResult['ignore-errors'] && !chksql($oDB->getRow('select * from place limit 1'))) { fail('No Data'); } } if ($aCMDResult['create-functions'] || $aCMDResult['all']) { - echo "Functions\n"; + info('Create Functions'); $bDidSomething = true; - if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built"); + if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) { + fail("nominatim module not built"); + } create_sql_functions($aCMDResult); } if ($aCMDResult['create-tables'] || $aCMDResult['all']) { + info('Create Tables'); $bDidSomething = true; - echo "Tables\n"; $sTemplate = file_get_contents(CONST_BasePath.'/sql/tables.sql'); $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); $sTemplate = replace_tablespace( @@ -230,12 +263,12 @@ if ($aCMDResult['create-tables'] || $aCMDResult['all']) { pgsqlRunScript($sTemplate, false); // re-run the functions - echo "Functions\n"; + info('Recreate Functions'); create_sql_functions($aCMDResult); } if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) { - echo "Partition Tables\n"; + info('Create Partition Tables'); $bDidSomething = true; $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql'); @@ -275,7 +308,7 @@ if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) { if ($aCMDResult['create-partition-functions'] || $aCMDResult['all']) { - echo "Partition Functions\n"; + info('Create Partition Functions'); $bDidSomething = true; $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql'); @@ -285,27 +318,25 @@ if ($aCMDResult['create-partition-functions'] || $aCMDResult['all']) { if ($aCMDResult['import-wikipedia-articles'] || $aCMDResult['all']) { $bDidSomething = true; - $sWikiArticlesFile = CONST_BasePath.'/data/wikipedia_article.sql.bin'; - $sWikiRedirectsFile = CONST_BasePath.'/data/wikipedia_redirect.sql.bin'; + $sWikiArticlesFile = CONST_Wikipedia_Data_Path.'/wikipedia_article.sql.bin'; + $sWikiRedirectsFile = CONST_Wikipedia_Data_Path.'/wikipedia_redirect.sql.bin'; if (file_exists($sWikiArticlesFile)) { - echo "Importing wikipedia articles..."; + info('Importing wikipedia articles'); pgsqlRunDropAndRestore($sWikiArticlesFile); - echo "...done\n"; } else { - echo "WARNING: wikipedia article dump file not found - places will have default importance\n"; + warn('wikipedia article dump file not found - places will have default importance'); } if (file_exists($sWikiRedirectsFile)) { - echo "Importing wikipedia redirects..."; + info('Importing wikipedia redirects'); pgsqlRunDropAndRestore($sWikiRedirectsFile); - echo "...done\n"; } else { - echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n"; + warn('wikipedia redirect dump file not found - some place importance values may be missing'); } } if ($aCMDResult['load-data'] || $aCMDResult['all']) { - echo "Drop old Data\n"; + info('Drop old Data'); $bDidSomething = true; $oDB =& getDB(); @@ -339,34 +370,45 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) { } // used by getorcreate_word_id to ignore frequent partial words - if (!pg_query($oDB->connection, 'CREATE OR REPLACE FUNCTION get_maxwordfreq() RETURNS integer AS $$ SELECT '.CONST_Max_Word_Frequency.' as maxwordfreq; $$ LANGUAGE SQL IMMUTABLE')) fail(pg_last_error($oDB->connection)); + $sSQL = 'CREATE OR REPLACE FUNCTION get_maxwordfreq() RETURNS integer AS '; + $sSQL .= '$$ SELECT '.CONST_Max_Word_Frequency.' as maxwordfreq; $$ LANGUAGE SQL IMMUTABLE'; + if (!pg_query($oDB->connection, $sSQL)) { + fail(pg_last_error($oDB->connection)); + } echo ".\n"; // pre-create the word list if (!$aCMDResult['disable-token-precalc']) { - echo "Loading word list\n"; + info('Loading word list'); pgsqlRunScriptFile(CONST_BasePath.'/data/words.sql'); } - echo "Load Data\n"; + info('Load Data'); + $sColumns = 'osm_type, osm_id, class, type, name, admin_level, address, extratags, geometry'; + $aDBInstances = array(); $iLoadThreads = max(1, $iInstances - 1); for ($i = 0; $i < $iLoadThreads; $i++) { $aDBInstances[$i] =& getDB(true); - $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, '; - $sSQL .= 'housenumber, street, addr_place, isin, postcode, country_code, extratags, '; - $sSQL .= 'geometry) select * from place where osm_id % '.$iLoadThreads.' = '.$i; - $sSQL .= " and not (class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString')"; + $sSQL = "INSERT INTO placex ($sColumns) SELECT $sColumns FROM place WHERE osm_id % $iLoadThreads = $i"; + $sSQL .= " and not (class='place' and type='houses' and osm_type='W'"; + $sSQL .= " and ST_GeometryType(geometry) = 'ST_LineString')"; + $sSQL .= " and ST_IsValid(geometry)"; if ($aCMDResult['verbose']) echo "$sSQL\n"; - if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection)); + if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) { + fail(pg_last_error($aDBInstances[$i]->connection)); + } } // last thread for interpolation lines $aDBInstances[$iLoadThreads] =& getDB(true); - $sSQL = 'select insert_osmline (osm_id, housenumber, street, addr_place, postcode, country_code, '; - $sSQL .= 'geometry) from place where '; + $sSQL = 'insert into location_property_osmline'; + $sSQL .= ' (osm_id, address, linegeo)'; + $sSQL .= ' SELECT osm_id, address, geometry from place where '; $sSQL .= "class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString'"; if ($aCMDResult['verbose']) echo "$sSQL\n"; - if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection)); + if (!pg_send_query($aDBInstances[$iLoadThreads]->connection, $sSQL)) { + fail(pg_last_error($aDBInstances[$iLoadThreads]->connection)); + } $bAnyBusy = true; while ($bAnyBusy) { @@ -378,11 +420,22 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) { echo '.'; } echo "\n"; - echo "Reanalysing database...\n"; + info('Reanalysing database'); pgsqlRunScript('ANALYSE'); + + $sDatabaseDate = getDatabaseDate($oDB); + pg_query($oDB->connection, 'TRUNCATE import_status'); + if ($sDatabaseDate === false) { + warn('could not determine database date.'); + } else { + $sSQL = "INSERT INTO import_status (lastimportdate) VALUES('".$sDatabaseDate."')"; + pg_query($oDB->connection, $sSQL); + echo "Latest data imported from $sDatabaseDate.\n"; + } } if ($aCMDResult['import-tiger-data']) { + info('Import Tiger data'); $bDidSomething = true; $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql'); @@ -440,7 +493,7 @@ if ($aCMDResult['import-tiger-data']) { echo "\n"; } - echo "Creating indexes\n"; + info('Creating indexes on Tiger data'); $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_finish.sql'); $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate); $sTemplate = replace_tablespace( @@ -457,116 +510,91 @@ if ($aCMDResult['import-tiger-data']) { } if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all']) { + info('Calculate Postcodes'); $bDidSomething = true; $oDB =& getDB(); - if (!pg_query($oDB->connection, 'DELETE from placex where osm_type=\'P\'')) fail(pg_last_error($oDB->connection)); - $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) "; - $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,calculated_country_code,"; - $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select calculated_country_code,postcode,"; - $sSQL .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y "; - $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x"; - if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection)); + if (!pg_query($oDB->connection, 'TRUNCATE location_postcode')) { + fail(pg_last_error($oDB->connection)); + } + + $sSQL = "INSERT INTO location_postcode"; + $sSQL .= " (place_id, indexed_status, country_code, postcode, geometry) "; + $sSQL .= "SELECT nextval('seq_place'), 1, country_code,"; + $sSQL .= " upper(trim (both ' ' from address->'postcode')) as pc,"; + $sSQL .= " ST_Centroid(ST_Collect(ST_Centroid(geometry)))"; + $sSQL .= " FROM placex"; + $sSQL .= " WHERE address ? 'postcode' AND address->'postcode' NOT SIMILAR TO '%(,|;)%'"; + $sSQL .= " AND geometry IS NOT null"; + $sSQL .= " GROUP BY country_code, pc"; + + if (!pg_query($oDB->connection, $sSQL)) { + fail(pg_last_error($oDB->connection)); + } if (CONST_Use_Extra_US_Postcodes) { - $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) "; - $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,'us',"; - $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode"; - if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection)); + // only add postcodes that are not yet available in OSM + $sSQL = "INSERT INTO location_postcode"; + $sSQL .= " (place_id, indexed_status, country_code, postcode, geometry) "; + $sSQL .= "SELECT nextval('seq_place'), 1, 'us', postcode,"; + $sSQL .= " ST_SetSRID(ST_Point(x,y),4326)"; + $sSQL .= " FROM us_postcode WHERE postcode NOT IN"; + $sSQL .= " (SELECT postcode FROM location_postcode"; + $sSQL .= " WHERE country_code = 'us')"; + } else { + $sSQL .= "TRUNCATE TABLE us_postcode"; } -} + if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection)); -if ($aCMDResult['osmosis-init'] || ($aCMDResult['all'] && !$aCMDResult['drop'])) { // no use doing osmosis-init when dropping update tables - $bDidSomething = true; - $oDB =& getDB(); + // add missing postcodes for GB (if available) + $sSQL = "INSERT INTO location_postcode"; + $sSQL .= " (place_id, indexed_status, country_code, postcode, geometry) "; + $sSQL .= "SELECT nextval('seq_place'), 1, 'gb', postcode, geometry"; + $sSQL .= " FROM gb_postcode WHERE postcode NOT IN"; + $sSQL .= " (SELECT postcode FROM location_postcode"; + $sSQL .= " WHERE country_code = 'gb')"; + if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection)); - if (!file_exists(CONST_Osmosis_Binary)) { - echo "Please download osmosis.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n"; - if (!$aCMDResult['all']) { - fail("osmosis not found in '".CONST_Osmosis_Binary."'"); - } - } else { - if (file_exists(CONST_InstallPath.'/settings/configuration.txt')) { - echo "settings/configuration.txt already exists\n"; - } else { - passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_InstallPath.'/settings'); - // update osmosis configuration.txt with our settings - passthru("sed -i 's!baseUrl=.*!baseUrl=".CONST_Replication_Url."!' ".CONST_InstallPath.'/settings/configuration.txt'); - passthru("sed -i 's:maxInterval = .*:maxInterval = ".CONST_Replication_MaxInterval.":' ".CONST_InstallPath.'/settings/configuration.txt'); + if (!$aCMDResult['all']) { + $sSQL = "DELETE FROM word WHERE class='place' and type='postcode'"; + $sSQL .= "and word NOT IN (SELECT postcode FROM location_postcode)"; + if (!pg_query($oDB->connection, $sSQL)) { + fail(pg_last_error($oDB->connection)); } + } + $sSQL = "SELECT count(getorcreate_postcode_id(v)) FROM "; + $sSQL .= "(SELECT distinct(postcode) as v FROM location_postcode) p"; - // Find the last node in the DB - $iLastOSMID = $oDB->getOne("select max(osm_id) from place where osm_type = 'N'"); - - // Lookup the timestamp that node was created (less 3 hours for margin for changsets to be closed) - $sLastNodeURL = 'http://www.openstreetmap.org/api/0.6/node/'.$iLastOSMID."/1"; - $sLastNodeXML = file_get_contents($sLastNodeURL); - preg_match('#timestamp="(([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}):([0-9]{2}):([0-9]{2})Z)"#', $sLastNodeXML, $aLastNodeDate); - $iLastNodeTimestamp = strtotime($aLastNodeDate[1]) - (3*60*60); - - // Search for the correct state file - uses file timestamps so need to sort by date descending - $sRepURL = CONST_Replication_Url."/"; - $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1"); - // download.geofabrik.de: 000/